measurements of cp asymmetries and branching fractions in b 0 p + p - , k + p - , k + k -

$: Measurements of CP asymmetries and branching fractions in B 0 p + p - , K + p - , K + K -$
16 May 2002 Paul Dauncey - BaBar 1

Measurements of CP asymmetries and branching fractions in B0 +,K+,K+K

Paul Dauncey

Imperial College, University of London

For the BaBar Collaboration


CP violation arises due to complex CKM matrix; e.g. Wolfenstein parametrisation:

)(O

1A)i1(A

A21

1

)i(A21

1

VVV

VVV

VVV

V 4

23

22

32

tbtstd

cbcscd

ubusud

is third internal angle

CP violation in the Standard Model

CKM unitarity; represent as triangle in , plane

Vtd e-i

Vub e-i


Direct CP violationCP violation requires interference between at least two amplitudes

with different phases, e.g. B0 K+

B0 K+

T

P

CP violation gives a non-zero asymmetry:

AK(B0 K(B0 K+(B0 K(B0 K+)

In principle leads tomeasurement

Relative

weak phase ei

– –


Indirect CP violationMixing gives two paths if final state accessible from B0 and B0, e.g.

B0 J/ K0S

B0 J/ K0S

CP violation gives a time-dependent asymmetry:

Amplitude = Im(ei) = sin2Leads tomeasurement: see talk by S. Rahatlou

B0

T

TM ––

–

Relative

weak phase ei2


CP violation in B0 +

B0 +has both effects

B0 +

If both tree and penguin significant: parametrise as sin2eff

Interpretation also depends on strong amplitudes

B0

T

T

PM

P

–

–

–Relative

weak phase ei

No relative weak phase

Relative

weak phase ei2


Experimental realityAnalysis is different from B0 J/ K0

S… No clean or K sample; need to determine particle type

Particle identification needed Analyse both modes simultaneously; also include KK

No clean signal sample; high backgrounds from udsc Branching fractions are ~ 10–6 – 10–5

Need to use kinematics and event shapes to distinguish modes Use unbinned maximum likelihood fit to extract signals

…but some elements are identical: Need to “tag” the other B to find if B0 or B0

Use lepton charge, K charge or neural networks Inefficiency and impurity (“dilution”)

Cannot measure time directly Time difference t between two B decays

–


The BaBar detector

DIRC PID)144 quartz bars

11000 PMs

1.5T solenoid EMC6580 CsI(Tl) crystals

Drift Chamber40 stereo layers

Instrumented Flux Returniron / RPCs (muon / neutral hadrons)

Silicon Vertex Tracker5 layers, double sided strips

e+ (3.1GeV)

e (9 GeV)

Pep-II delivers boosted e+e Y(4s) BB, on and off the Y(4s)

Integrated luminosity: 55.6 fb-1 on peak, 60.2 0.7 million BB events

–

–


Particle identification with the DIRCDetector of Internally Reflected Cherenkov light (DIRC): essential

for this analysis to distinguish K from Reconstruct Cherenkov

angle cosC = 1/n

from rings seen in PMTs

Cherenkov light transmitted down quartz bars by internal reflection


Analysis is done in two stages: Direct CP; extract branching fractions for K and KK and

also the K decay CP asymmetry AK

Maximum likelihood fit to kinematic/event shape quantities Requires no tag or vertex measurement Separate fit reduces systematic error

Indirect CP; extract CP asymmetry sin2eff

Fix branching fractions and AK to above results Requires tag to determine if B0 or B0

Requires vertex information to find time dependence

All results are preliminary

Analysis overview

–


B candidate mass using beam “energy substitution”

mES = (Ebeam2 – pB

2) in CM

Select 5.2 < mES < 5.3 GeV/c2

Kinematic variables - mES

Signal Monte Carlo; Gaussian

Background; ARGUS threshold function

Depends on reconstructed momentum of B candidate, pB ~ 325 MeV

Same for all signal modes Resolution dominated by

beam energy uncertainty mES) ~ 2.6 MeV/c2

Signal shape from MC, background from fit


Kinematic variables - E

Background; quadratic function

KKK

B candidate energy difference from beam energy

E = EB – Ebeam in CM

Select | E | < 0.15 GeV

Depends on masses of B decay products; mass assumed

K and KK shifted to non-zero average; ~ 45 MeV per K

Resolution dominated by tracking

E) ~ 26 MeV Signal shape parametrised from

MC; background from fit

Signals: Gaussian


DIRC c mean and resolution parametrised from data using D*+ D0+ (K–+)+ decays

Particle identification - c

(C) = 2.2 mrad 9

2.5

K/ Separation

Momentum range of decays


Background suppression - Fisher

Signal Monte Carlo

Background

Cut on angle between B candidate and sphericity of the other tracks in the event: |cos s| < 0.8

F - optimized linear combination of energy flow into nine cones around candidate (CLEO Fisher discriminant). Signal shape from MC, background from fit

Cut

Signal Monte Carlo

Background


Five input variables per event; assume independent (uncorrelated) PDFs for each:

mES

F Discriminate signal from background E C

+ Discriminate different signal modes

C–

Eight fit parameters: four for signal, four for background N(+ N(K+ N(+K N(K+ K

Branching fraction maximum likelihood fit

Fit directly for N(K) and AK

N(K+N(K) (1–AK

N(+KN(K) (1+AK


Branching fraction results

Mode

Yield

(events)

Branching Fraction

(10-6)

KAsymmetry, AK

πKB0

ππB0

KKB0

01.006.005.0

4.07.04.5

8.01.18.17

C.L.) (90% 1.1

716915124

1524403 C.L.) (90% 6.15

No significant direct CP violation seen in B0 K+

90% C.L. –0.14 AK 0.05

Main systematics:

Branching fractions – uncertainty in shape of C PDF

Asymmetry - possible charge bias in track and C reconstruction

Fit results are:


Branching fraction projections

Fit sample: 17585 candidates, ~ 38%

B0K

B0

Likelihood projections: remove one variable from fit and cut on fit probabilities to give enhanced signal samples:

Background and crossfeed


Measuring indirect CP violationExtend the branching fraction analysis to extract indirect CP

information: Needs extra information on:

Time of decay (vertexing); reconstruct “other” B decay point and find time difference of B decays

Flavour of the decaying B0 (tagging); use tracks whose charge carries flavour information

Use identical techniques to sin2 analysis

Fit for CP violation asymmetries while holding branching fractions and K asymmetry to previously determined values Vary these parameters within determined errors for systematics; small

effect for asymmetries


Asymmetric beam energies mean Y(4s) is boosted:Vertexing and t

(4s)

= 0.55 +z

t z/c

-

Fully reconstructed BBest vertex for “other” B

t is time difference between the two decays: Resolution is dominated by the “other” B, (t) ~ 0.8 ps Independent of signal type Can use same resolution model as sin2 analysis Exploit mixing to measure signal performance (dilutions) and

efficiencies

Signal resolution determined from large sample of fully reconstructed B’s, background shape determined from fit


The signal modes depend differently on t: general form allows for both tree and penguin: rates

f(t) ~ 1 Ssin(mdt) Ccos(mdt) for B0(B0) tag

SImand C For pure tree, ei2so Ssinand C With some penguin contribution,CandSC

sineff

Ktime dependence due to B0 mixing: rates f(t) ~ 1 cos(mdt) for unmixed (mixed) B0

KK general form similar to in principle Model with simple exponential

Actual PDFs used in fit: Diluted due to mistags Convolved with t resolution functions

Signal yield time dependences

–


Flavour tagging

Gev/c2mES

Fit sample: 9220 tagged candidates

Separate into tag categories

Lepton; eff ~ 8%

Neural Net 1; eff ~ 2%

Kaon;

eff ~ 14%

Neural Net 2; eff ~ 1%

Use same flavour tagging as sin2 measurement:

Untagged events (~ 33%) retained in the fit to determine background shapes


Fit results in:

S = –0.01 ± 0.37 ± 0.07

C= –0.02 ± 0.29 ± 0.07

No significant indirect CP violation seen in B0 +

90% C.L. –0.66 S 0.62

90% C.L. –0.54 C 0.48

Main systematic: Uncertainty in shape of C PDF

Time-dependent fit results

Enhanced B sample: t distributions and asymmetry between mixed and unmixed events.

Fit result

Fitted background


Fit checked using “toy” studies of simulated experiments All parameters unbiased All errors consistent with expectations Likelihood value (goodness of fit) consistent with expectations

Fit holds B lifetime and mixing constant: cross check by fitting B = 1.66 ± 0.09 ps (c.f. PDG value 1.54 ± 0.02 ps)

md = 0.517 ± 0.062 ps–1 (c.f. PDG value 0.479 ± 0.012 ps–1)

Time-dependent fit cross checks

Enhanced B Ksample: asymmetry between unmixed and mixed events.


Summary of BaBar preliminary results

Branching fractions:

B(B0 ) = (5.4 ± 0.7 ± 0.4) 10–6

B(B0 K) = (17.8 ± 1.1 ± 0.8) 10–6

B(B0 KK)= 1.1 10–6 (90% C.L.)

CP asymmetries; no evidence for CP violation:

S= –0.01 ± 0.37 ± 0.07 or 90% C.L. –0.66 S 0.62

C= –0.02 ± 0.29 ± 0.07 or 90% C.L. –0.54 C 0.48

AK= –0.05 ± 0.06 ± 0.01 or 90% C.L. –0.14 AK 0.05

measurements of cp asymmetries and branching fractions in b 0 p + p - , k + p - , k + k -

Documents

b0 p pb0 p p

p gb0

p gb0

effectsb0p p

pb0k ptpcp violation

b0 jy k0sb0jy k0scp

b0 jy k0sno clean pp

measurements of cp asymmetries